This invention relates to a scanning probe microscope for observing the surface geometry of a specimen, and more particularly to a scanning probe microscope suitable for measuring the roughness and tilt angle of the sidewall of a step in an IC semiconductor, for example, the sidewall of an electrode line pattern.
A scanning probe microscope (SPM) is an instrument that brings a probe as close as less than 1 .mu.m to the surface of a specimen, scans in the directions of X and Y or of X, Y, and Z while sensing the interaction between the probe and the specimen, and provides a two-dimensional mapping of the interaction. The scanning probe microscope (SPM) is the generic name for the scanning tunneling microscope (STM), the interatomic force microscope (AFM), the magnetic force microscope (MFM), the scanning near-field optical microscope (SNOM), etc. Of these, the AFM is the most widely used type of SPM as an instrument for acquiring information on the unevenness of the surface of a specimen. The AFM indirectly acquires information on the unevenness of the surface of a specimen by sensing the displacement of a cantilever caused by the force acting on a probe when the probe (a projection) formed at the tip of the cantilever is brought closer to the surface of the specimen.
The probe used for the AFM is produced using semiconductor processes, batch fabrication techniques, because of advantages in performance and cost. For example, what is called a flat lever produced by patterning a silicon oxide thin film has been reported in T. R. Albrecht et al., "Atomic Resolution with Atomic Force Microscopy," Europhys. Lett., 3(12), pp. 1281-1286, 1987. A bird's beak-type probe chip, an improved version of the flat lever, has been disclosed in Jpn. Pat. Appln. KOKAI Publication No. 1-262403. The silicon nitride probe chip with a pyramidal probe disclosed in U.S. Pat. No. 5,399,232 and the silicon probe chip disclosed in U.S. Pat. No. 5,051,379 have already been commercialized and available on the market. Such a probe chip uses the projection point-terminated at the tip of the probe as a virtual probe, with the vertical angle of the probe in the range from 15.degree. to 90.degree. in view of the whole probe.
As for semiconductor IC design rules, trial products of 256-MB storage device adhering to the 0.25 .mu.m rules have been developed and the 0.15 .mu.m rules are going to be applied to one-GB storage devices. Such being the case, the element geometry inspecting machine is required to accurately measure the width of narrow lines of an element with a large aspect ratio and the entire geometry. Because scanning probe microscopes are suitable for such geometry measurement, they have been under intensive study.
In Yyes Martin and H. Kumar Wickramasinghe, "Method for imaging sidewalls by atomic force microscopy," Apply. Phys. Lett. Vol. 64(19), pp. 2489-2500, 1994, a new scanning probe microscope that images vertical walls have been proposed. A patent related to this has been disclosed in Japanese Patent No. 2,501,282. With the scanning probe microscope, use of a boot-type probe (i.e., a cylindrical probe whose body is narrow near the tip) enables the measurement of the vertical wall of a specimen. Unlike the aforementioned point-terminated probe, such a probe allows different points at the flare portion of the tip to interact with the sidewalls on both sides of the recessed portion. Namely, the boot-type probe has at least two or more virtual probes at the tip, which makes the vertical angle of the virtual probe 0.degree. or less.
A probe of the boot type is composed of a portion 2 .mu.m to 2.5 .mu.m in diameter on the cantilever side (a thick portion) and a thin portion connected to the tip of the thick portion. The thin portion is shaped like a boot. The boot-shaped portion has a length (height) of 2.8 .mu.m, the probe at the tip has a diameter of 360 nm at the flare section, and the narrow portion closer to the cantilever has a diameter of 210 nm.
When the sidewalls of a trench or a hole in a semiconductor are measured using the probe chip, it is the flare section that comes closest to the surface of the specimen (sidewalls), since the flare section at the tip of the probe projects. The roughness or tilt angle of the sidewall can be measured by causing the probe to scan the specimen while keeping the spacing between the projecting portion of the probe and the surface of the specimen constant.
Jpn. Pat. Appln. KOKAI Publication No. 3-104136 has disclosed a method of producing such a boot-type probe. The probe is produced by photolithography using a monocrystalline silicon wafer as a start wafer. After a circular mask with a diameter of about 1 .mu.m or less has been formed, the silicon wafer is dug almost vertically by dry etching with CF4 gas to form a semi-cylindrical silicon probe section. Changing the conditions for dry etching makes the semi-cylindrical probe section thicker or thinner in the middle of the cylinder. Suitably selecting the conditions produces a monocrystalline semi-cylindrical probe thinner in the middle. Thereafter, the probe section is protected with resist or the like. The patterning of the cantilever is achieved, followed by etching from the reverse side of the wafer. This produces a probe chip with a boot-type probe.
The probe of the AFM probe chip, however, can permit the tip of the probe to wear or break because it is in contact with the surface of the specimen during measurement (scanning). Attention should be given to a material for the probe to make stable AFM measurements. For example, Matsuyama, et al., gave a presentation on wear of probe materials in an academic lecture in the 55th Applied Physics Society Meeting (proceedings p. 473). They reported that monocrystalline silicon and silicon nitride were widely used materials, comparison between them showed that a silicon nitride film wears less than a monocrystalline silicon film, and a silicon nitride film with a silicon-nitride stoichiometry of 3:4 wears still less among silicon nitride films.
Not to mention a scanning probe microscope that images vertical walls by Yves Martin, et al., even usual scanning probe microscopes need a probe whose vertical angle is small and whose aspect ratio is high to measure the root of the sidewall of a specimen with a high aspect ratio as accurately as possible.
A scanning probe microscopy of imaging vertical walls by Yves Martin, et al., described in Apply. Phys. Lett. Vol. 64(19), pp. 2489-2500, 1994, is a method using noncontact mode AFM. During measurement in the air, the probe sometimes comes into contact with the surface of the specimen. Specifically, although the noncontact mode AFM measuring method is used, the finite band of the feedback circuit in the system makes it difficult to measure a specimen with great irregularities or a specimen with steps without allowing the probe to touch the specimen at all. The difficulty of measuring vertical walls by the complete noncontact mode AFM measuring method can be inferred from the fact that Yves Martin, et al., authors of the thesis, have disclosed in Jpn. Pat. Appln. KOKAI Publication No. 6-19415 a method of measuring the sidewalls of a specimen with similar steps by a contact mode AFM method with the probe being excited and from the fact that Virgil B. Elings, et al., has disclosed in Jpn. Pat. Appln. KOKAI Publication No. 7-270434 a microscope capable of measuring sidewalls by a contact mode AFM method with the probe being exited.
When the probe comes into contact with the specimen, this possibly permits the probe to wear or break, changing the shape of the probe during the measurement. Because the scanning probe microscopy of imaging vertical walls by Yves Martin, et al., is used to measure the shapes of electron patterns of semiconductor ICs complying with submicron pattern rules as described above, even when the probe has worn and its shape has changed as little as several tens of nanometers, the measurement result changes 10% or more. Therefore, the change of the shape of the probe due to wear becomes a very serious problem. In general, measuring devices are required to have a high reproducibility. When the scanning probe microscope that images vertical walls by Yves Martin, et al., is considered in terms of measuring devices, the fact that wear of the probe will ruin the reproducibility of measurement data is a very serious problem.
To assure the reproducibility of data, frequent correction of the shape of the probe can be considered. Frequent correction of the probe before measurement of a specimen raises a problem in terms of throughput required of measuring devices. The reason is that the number of specimens that can be measured during a unit time decreases because of correction work. If a correction method of measuring a correction specimen by the same method as in measurement is used for correction, the probe can wear during correction work and its shape change. Therefore, too frequent correction work would cause a problem.
Furthermore, the flare portion projects from the tip of the boot-type probe for use with a scanning probe microscope that images vertical walls, which has been disclosed by Yves Martin, et al. The projection permits at least two virtual probes to exist, but it is difficult to determine which part of the flare portion is interacting with the specimen.
Furthermore, a probe used in the scanning probe microscopy that images vertical walls written in the thesis by Yves Martin, et al., is formed by etching a monocrystalline silicon as disclosed in Jpn. Pat. Appln. KOKAI Publication No. 3-104136. As described above, silicon is a material liable to wear. Wear is a serious problem in terms of probe material.
To form a monocrystalline silicon boot-shaped probe disclosed in Apply. Phys. Lett. Vol. 64(19). pp. 2489-2500, 1994 or Jpn. Pat. Appln. KOKAI Publication No. 3-104136, dry etching has been used. In general, when dry etching is used for submicron patterning, a slight change in the formation conditions causes the problem of changing the shape of the probe.
To measure a specimen of the submicron order, it is necessary to use a probe narrower than that dimension. It is quite difficult to form such a probe with good uniformity. The shape varies depending on not only the wafer but also the place in one wafer. From this, it can be understood easily how difficult it is to form a probe narrower than dimensions of the submicron order using a machine similar to a dry etching machine for manufacturing semiconductor ICs adhering to pattern rules of the submicron order.
Variations in the shape (dimensions) of the probe lead to a rise in the cost of manufacturing probes, which becomes a serious problem. Specifically, to measure the submicroscopic irregularities in a specimen, it is necessary to examine the probes carefully before shipment whether each probe is finished to dimensions smaller than at least the design dimensions. This raises the inspection cost. Variations in the shape of the probe, of course, lowers the yield considerably, resulting in a further rise in the cost.